Diagnostic accuracy of late iodine-enhancement dual-energy computed tomography for the detection of chronic myocardial infarction compared with late gadolinium-enhancement 3-T magnetic resonance imaging.

OBJECTIVES: The purpose of the study was to compare the performance of late iodine-enhancement (LIE) dual-energy computed tomography (DECT) linear blending and selective myocardial iodine mapping for the detection of chronic myocardial infarction (CMI) with late gadolinium-enhancement (LGE) 3-T magnetic resonance imaging.

MATERIALS AND METHODS: This study was approved by the institutional review board, and the patients gave informed consent. A total of 20 patients with a history of CMI underwent cardiac LIE-DECT and LGE-MRI. Images of the LIE-DECT were reconstructed as 100 kilovolt (peak) (kV[p]), 140 kV(p), and weighted-average (WA; linear blending) images from low- and high-kilovoltage peak data using 3 different weighting factors (0.8, 0.6, 0.3). Additional color-coded myocardial iodine distribution maps were calculated. The images were reviewed for the presence of late enhancement, transmural extent, signal characteristics, infarct volume, and subjective image quality.

RESULTS: Segmental analysis of LIE-DECT data from 100 kV(p), WA of 0.8, and WA of 0.6 showed identical results for the identification of CMI (89% sensitivity, 98% specificity, 96% accuracy) and correctly identified all segments with transmural scarring detected through LGE-MRI. Weighted average of 0.6 received the best subjective image quality rating (15/20 votes) and average measured infarct size correlated best with LGE-MRI (5.7% difference). In comparison with LGE-MRI, iodine distribution maps were susceptible to false-positive and false-negative findings (52% sensitivity, 88% specificity, 81% accuracy), overestimating quantity of transmural scars by 78% while underestimating infarct volume by 55%.

CONCLUSIONS: Late iodine enhancement cardiac dual-energy computed tomography correlates well with LGE-MRI for detecting CMI, whereas iodine distribution analysis provides inferior accuracy. Linear blending further improves image quality and enables more precise estimation of scar volume.